Bias: Error from oversimplified assumptions. High bias = underfitting. Model can't capture the pattern.
Variance: Error from sensitivity to training data fluctuations. High variance = overfitting. Model memorizes noise.
The tradeoff: Simple models have high bias, low variance. Complex models have low bias, high variance. Goal: find the sweet spot.
Interview question: "Your model has high training accuracy but low test accuracy. What's wrong?"
Overfitting (high variance). Solutions: more data, regularization, simpler model, dropout, early stopping.